I found the reason for this bug. Turn OFF “Persistent Data” and the bug is gone (as a workaround). After this you will be able to change the render-sample-rate
Version of today (28 Jun) don’t recognize my 3090 as an OptiX GPU anymore (Win 10, 462.29 driver). Have to fall back to Cuda. Had this problem a few versions back (1-2 weeks) then it got fixed. Please find and exterminate this bug.
Hi, there is no bug fixing for this build until it is merged with the master.
I always use the .zip version and keep a working version.
One mention to use at least driver 470, cant remember if it was in this thread or the BA thread.
Cheers, mib
Installing 471.41 studio driver fixed it. Optix enabled again. Just Blenders way of subtle remainding to update my drivers. Cheers.
Getting greenish denoising with OptiX denoising on my 3090 in Optix in the Viewport with the 28 Jul build.
Getting less green with more samples. Have this for a few weeks now, even with newest 471.41 studio driver. Looks like a denoiser debug sample display setting left on. EDIT: greenish “error” only with “Color + Albedo + Normal” input passes setting.
This has been going on for a while, seems to be including the normal colors in the denoising final image.
2 Likes
Adaptive sampling on volumes renders longer with very noisy volumes.Very noisy. No matter how smallish the threshold, even on < 0.00001. On the plus side volume rendering without adaptive render a lot cleaner with relatively low samples since the Equiangular volume sampling update.
Still having this strange motionblur bug on rotating geonetry with Optix with the 28 Jul build.
Just watched the first few minutes, I am not intelligent enough to understand the content so I stopped. But I do hear the words “spectral” “wavelength” or something a lot in the video, I guess this will be relevant to the Spectral Cycles project? I am not sure, this video is long and full with math equations, my impression might be completely off on what this is about 
1 Like
Is OIDN 1.4.1 already in pre-built libraries (Linux)? “config.h” file mentions 1.4.1 but I’m not sure if this is just a definition or something.
I just want to know to apply Sergey patch on Cycles X branch for testing.
https://developer.blender.org/D12043
Hi YAFU, I build myself and iirc the libOpenImageDenoise.a was updated yesterday.
Author: Sybren A. Stüvel <[email protected]>
Date: Tue Jul 27 10:15:31 2021 +0200
Deps builder: OIIO/OSL/ISPC/OIDN/LLVM/Flex updates
This diff somewhat snowballed out of updating OIDN to 1.4.1 it had some
changes that allowed us to remove the arm hacks we had in place and
revert to using identical versions for a whole bunch of deps. But that
required an update to ISPC which needed a newer LLVM and if we're
updating LLVM we may as well update OSL, and when we update OSL, OIIO
may as well be dragged in soo......anyhow...
This diff updates:
LLVM 9.0.0 (11.0.1 for mac/arm) -> 12.0.0
OIIO 2.1.15.0 -> 2.2.15.1
OSL 1.11.10.0 -> 1.11.14.1
winflex_bison 2.5.5-> 2.5.24 (ispc needed newer bison, windows only dep)
OIDN 1.4.0 -> 1.4.1
ISPC v1.14.1(random hash for mac/arm) -> v1.16.0
Flex 2.6.4 (ISPC needed newer Flex than available on CentOS 7)
Cheers, mib
2 Likes
Basically, the video highlights that a correct Fresnel value is imperceptible in the rendering engines for film production, towards the end of the video there are various comparisons, with and without Fresnel.
I’m testing the patch, “fishy cat” scene at 10 samples. There is noticeable noise in the cat’s eyes when Prefilter is enabled. In the View Layers tab, Denoising, Input Passes: Color only there is no noise, Color+Albedo the noise is more noticeable and Color+Albedo+Normal the noise is even more noticeable.
So I’m not sure if this is simply due to some feature that is missing from Cycles-X.
Hi, the patch is now in the Cycles-x branch.
Cheers, mib
1 Like
I also noticed issues like that, you can see my comment about it here: ⚙ D12043 Cycles X: Support OIDN with guiding pass prefiltering
From the looks of it, the CleanAux function seems to have the largest impact on the visual quality of the image while denoising with OIDN. This function is now enabled with every OIDN render.
Pre-filtering should help reduce this noise, or get rid of it entirely, however I could not observe a difference between pre-filtering on and off, hence I believe it may not be working properly.
2 Likes
In Fishy Cat scene the Denoising Albedo pass is still noisy in eyes.
In the first implementation I think that I did not notice differences either. With yesterday’s patch there are differences, for example in this scene:
https://pasteall.org/blend/0cd5dd27d96e4a96a3604238221f0228
Prefilter OFF:
Prefilter ON:
But with Prefilter enabled I get noisy result where Denoising Albedo pass is noisy in Cycles-X, for example in Fishy Cat eyes, or in metal parts of the chairs in Classroom scene (at low samples).
I will test later with today’s build (when new Cycles-X buildbot experimental build is available).
I have built from Cycles X branch. Now Prefilter option seems to be inverted/reversed? Apparently with the Prefilter box checked, I get what I got before with the Prefilter box unchecked.
(Try with the scene that I have published above with samples=1 to see it better)
So, now texture is better preserved with Prefilter option unchecked than with Prefilter option checked.
There has been a fix today, perhaps it’s working fine now (haven’t tested myself)
https://developer.blender.org/rBfd9c4f166b61ac91db792c1d6b637f06faac902b
Related to the problem that I had mentioned above related to the option would seem to be inverted, I think that my build includes that commit.
Cycles X new version fd9c4f166b61 Cycles crashing when entering rendering in the viewport.